The prognostic influence of body mass index, resting energy expenditure and fasting blood glucose on postoperative patients with esophageal cancer

Background Body mass index (BMI), resting energy expenditure (REE) and fasting blood glucose (FBG) are major preoperative assessments of patients’ nutrition and metabolic state. The relations and effects of these indices on esophageal cancer patients’ postoperative short-term and long-term outcomes remain controversial and unclear. We aimed to study the impact of BMI, REE and FBG in esophageal cancer patients undergoing esophagectomy. Methods Three hundred and six esophageal cancer patients who underwent esophagectomy were observed retrospectively. Clinical characteristics, postoperative complications and survival analysis were compared among different BMI, REE and FBG groups. Results There were significant linear relationships between REE, BMI and FBG indices, patients with low BMI tended to have low REE (p < 0.001) and low FBG (p = 0.003). No significant difference was found in case of mortality and postoperative complications among different groups. Low BMI (X 2 = 6.141, p = 0.046), REE (X 2 = 6.630, p = 0.010) and FBG (X 2 = 5.379, p = 0.020) were related to poor survival. FBG ≤90 mg/dL was independently associated with poor survival (HR = 0.695; 95 % CI 0.489–0.987, p = 0.042). BMI and REE came to be stronger prognostic factors on lymph node-negative patients and proved to be independent prognostic indicators (HR = 0.540; 95 % CI 0.304–0.959, p = 0.035 and HR = 0.457; 95 % CI 0.216–0.967, p = 0.041, respectively). Conclusions BMI, REE and FBG are important prognostic factors in patients with esophageal cancer undergoing esophagectomy and preoperative evaluation of these indices help to determine the prognosis in these patients.


Background
Esophageal cancer (EC) is the eighth most common cancer and the sixth leading cause of cancer mortality worldwide [1]. Esophageal adencarcinoma (EAC) and esophageal squamous cell carcinoma (ESCC) are the most frequent histological subtypes. ESCC is the dominant histological subtype in china [2]. The potential prognostic indicators of esophageal cancer include histological variants (histological grading, differentiation, invasion depth and classification of lymph node metastasis) and nutrition or inflammation based prognostic factors (total lymphocyte counts, neutrophil to lymphocyte ratio (NLR), serum albumin and so on). BMI, REE and FBG, which are widely used to assess preoperative nutritional and metabolic status, have also been described to be prognostic predictors in several tumors [3].
Previous studies reported that high BMI was associated with increased risk of EAC while low BMI with ESCC [4]. High BMI in surgical patients is thought to be associated with increased comorbidities and postoperative complications, but the influence of high BMI on survival in patients undergoing esophagectomy is controversial. Hayashi found patients with high BMI showed better overall survival (OS) and disease-free survival (DFS) because of the early clinical diagnosis [5]. In contrast to this, Yoon pointed out that high BMI was independently associated with two-fold worsening of DFS, and OS after surgery for EAC [6]. Blom concluded that BMI had no prognostic value on short-term and longterm outcomes [7]. Compared with the previous studies, the mean BMI in Western populations was higher and had a greater incidence rate of EAC whereas the mean BMI in Chinese population was found to be lower with predominant ESCC. The effect of low BMI on postoperative complications and long-term survival remains unclear. So our study assessed the relationship between BMI and postoperative complications and long-term survival in Chinese population.
Disorder of energy metabolism is a common phenomenon in cancer patients. The energy metabolic status among different cancers may not be the same. Many researchers assessed the resting energy expenditure on cancer patients and no unanimous conclusions have been drawn. In this study, patients' preoperative REE was an estimated variable, and calculated by the Mifflin-St. Jeor equation [8]. REE is the sum of the metabolic activities of internal body and can reflect patient's physiques and muscle volumes, and REE per kg total body weight (REE/kg) may reflect the diffusion and metabolic rate of muscle more accurately. So both REE and REE/kg were used to explore its effects on short-term and long-term outcomes on esophageal cancer patients.
Epidemiologic evidence suggests that elevated blood glucose is associated with many forms of cancer [9]. However, opinions toward the association between elevated blood glucose and outcomes are mixed. Elevated blood glucose may promote cancer progression and lead to poor outcomes via pathways mediated high levels of insulin and insulin-like growth factor [10]. On the other hand, some studies proposed that diabetes-related microvessel changes might play a protective role against "neoplastic" cell metastasis in cancer patients and enhance cancer prognosis [11]. For EC, the relationship between the circulating glucose levels and prognosis has never been reported. To examine effects of fasting glucose levels at time of cancer diagnosis on postoperative outcomes is another aim of this study.
So the purpose of the study is to investigate the prognostic value using combined assessment of BMI, REE and FBG in esophageal cancer patients.

Patients
Patients who underwent esophagectomy and lymph node dissections during Sep 1, 2003 to Dec 31, 2008 were observed. Twenty-three patients who had palliative or R1/R2 resections were excluded. Patients who had histology other than SCC or AC (one lymphoma, one melanoma, one neuroendocrine carcinoma, one stromal tumor and two small cell carcinoma) were excluded. Patients who received neoadjuvant therapy were also excluded. Three hundred and six patients with histologic documentation of AC or SCC were included. All patients provided written consent and this retrospective study was approved by the Institutional Review Board of Huashan Hospital, Fudan University.

Clinical data
Preoperative staging was performed in all patients by means of barium meal test, fibro-gastroscopy, computed tomography of the chest and abdomen and ultrasound of the neck and retroperitoneal lymph nodes. Some patients also received whole body positron emission tomography (PET)-CT scanning. All patients were assessed for physiological ability to undergo esophagectomy. These evaluations included pulmonary function test, cardiac function test and nutritional assessment.
One hundred and eighty-six patients with tumors in middle or lower thoracic esophagus and no evidence of lymph node involvement in the superior mediastinum or neck region received transthoracic esophagectomy via left thoracotomy which also included a two-field lymphadenectomy; 53 patients with tumors in the middle or upper thoracic esophagus and possible lymph node metastasis in the superior mediastinum or neck region received McKeown 3-hole esophagectomy with three-field lymphadenectomy; 67 patients with tumors in the esophagogastric junction received esophagogastrectomy through median laparotomy.
TNM staging was performed according to the AJCC 7th edition guidelines. Patients were followed up every 6 months for the first 3 years and then annually. Survival time was measured as the time from the date of surgery to the date of death or the latest follow-up time. Endoscopy, CT, PET-CT and Radionuclide bone scan were performed if recurrent or metastatic disease was suspected.
Patients' weight and height were measured at their first hospitalization. BMI was calculated as weight in kilograms divided by height in meters squared.
The serum FBG concentration of patients was measured in the morning during the first hospitalization by hexokinase method after fasting for 10 h. Serum albumin was measured by means of the bromocresol purple method using automated equipments, and all patients were screened and excluded from acute and chronic liver disease. The complete blood count test (leukocyte, neutrophil, lymphocyte, monocyte, eosinophil, basophil, platelet counts, and hemoglobin) was carried out by an automated haematology analyser within one week prior to surgery. NLR was calculated as the ratio of neutrophils to lymphocytes in peripheral blood. These reagents and equipments were convenience-validated and standardized in our central clinical laboratory.
The 6-month preoperative weight loss was also measured and patients were divided into three categories: No/Little (loss of 0 to 5 % body weight), Middle (loss of 5 to 10 % body weight) and Large (loss more than 10 % body weight).
During the first few days after surgery, patients were treated with total parenteral nutrition. An initial dose of 5-10 kcal.day −1. kg −1 of enteral nutrition was supplied via a duodenum or jejunum feeding tube from the 2nd or 3rd postoperative day and gradually increased to the full dose of 25-30 kcal.day −1. kg −1 . Some patients with low BMI or low serum albumin levels were treated with full dose of enteral nutrition from one week before surgery to the tenth day after surgery.

Statistical analysis
All statistical analysis was performed with SPSS 16.0. Spearman rank order correlation coefficient (Spearman's rho) for nonparametric data was used. Univariate analysis of survival was performed using Kaplan-Meier method and log-rank test to estimate the prognostic value. Multivariate analysis of survival was performed using Cox-regression model to estimate hazard ratios (HRs) with 95 % confidence intervals (CI) and identify independent prognostic factors. The level of significance was set to p < 0.05.

Patient characteristics
The clinical characteristics and 5-years survival rate are summarized in Table 1. The median follow-up time was 37 months. The overall 1-, 3-and 5-year survival rate was 86.6, 60.8 and 47.1 % respectively.

Patient characteristics by BMI, REE, REE/kg and FBG
As shown in Table 2. Spearman's rho indicated that patients with age > 65 years (p < 0.001) and advanced Tstage (p = 0.040) were more likely to fall in the low REE class. Smoking patients were more likely to be associated with low BMI class (p = 0.013), low FBG class (p = 0.039) but high REE/kg class (p = 0.001).
It was interesting to explore the relationship between REE and REE/kg. Both high REE and low REE/kg patients were found to have better survival, which was consistent with the spearman correlation analysis that found patients with high REE tended to have lower REE/ kg (p < 0.001).
There were significant linear relations between BMI, REE, and FBG, patients with low BMI tended to have low REE (p < 0.001) and low FBG (p = 0.003). The trend of a linear association between REE and FBG could also be seen (p = 0.011).
There were close relations between REE/kg and weight lost or BMI, patients with high REE/kg have been found to lose more weight (p = 0.006) and have lower BMI (p < 0.001).
Low FBG was more likely to be seen in patients with high REE/kg (p = 0.025) and low albumin (p < 0.001).

Comorbidities, postoperative mortality and postoperative complications
Preoperative comorbidities, mortality and postoperative complications among different groups of BMI, REE, REE/kg and FBG were presented in Table 2. COPD appeared to be more frequent in low REE patients (p = 0.022), and diabetes was more common in high BMI patients (p = 0.006). All the seven patients with cardiovascular disease belonged to low FBG class (p < 0.001). Observing the short-term outcomes, no significant difference was found in postoperative mortality and major postoperative complications among different BMI, REE, REE/kg and FBG groups.

Nutrition or inflammation-based prognostic factors
Univariate analysis of nutrition or inflammation-based prognostic factors found to be serum albumin (p < 0.001), but not Total lymphocyte counts (p = 0.780) or NLR (p = 0.787), which was positively related to survival. The spearman correlation analysis found higher serum albumin levels were observed in high BMI, REE and FBG but in low REE/kg groups.

Univariate, multivariate and subgroup analysis
To evaluate the prognostic factors potentially related to survival, univariate analysis was applied (Table 1) and found no statistical associations of histologic subtype, surgical type, weight loss and adjuvant chemoradiation with OS. The prognostic factors were age, sex, differentiation, T-stage, N-stage, BMI (X 2 = 6.141, p = 0.046, Fig. 1 In order to assess the impact of BMI and REE on different tumor stages, we divided patients into subgroups on the basis of N-stage (N0 vs. N1-4). Patients with higher BMI and REE had significantly better OS in subgroup of N0 (X 2 = 15.507 p < 0.001 and X 2 = 14.717, p < 0.001, respectively), but no statistical significance in subgroup of N1-4 (X 2 = 1.952, P = 0.377 and X 2 = 0.386, p = 0.534, respectively). Multivariate analysis of subgroups revealed BMI (HR = 0.540; 95 % CI 0.304-0.959, p = 0.035) and REE (HR = 0.457; 95 % CI 0.216-0.967, p = 0.041) were independent prognostic factors for N0 patients but not for N1-4 patients.

Discussion
Our study identified that FBG level ≤90 mg/dL was independently associated with poor survival and also confirmed that advanced cancer stages remain the most powerful prognostic factors. In addition, we observed that BMI, REE and FBG were significant prognostic factors, but the prognostic value of BMI and REE is not the same between different lymph node metastasis statuses. For metastatic esophageal cancers, the most important prognostic factors were FBG, albumin and cancer staging including differentiation and invasion depth    whereas BMI and REE did not significantly affect the OS. However, for non-metastatic esophageal cancers, BMI and REE were important risk factors and proved to be independent prognostic indicators. Total lymphocyte counts, NLR and serum albumin were recognized as nutrition based or inflammation-based prognostic factors. Among the cancer patients, those with digestive tract malignancies were more likely to suffer from hypoalbuminaemia, which has been attributed to increased catabolism, obstruction of the digestive tract and the systemic inflammatory response [12]. The combined use of albumin and serum C-reactive protein  were introduced by the Glasgow prognostic score [13,14] and the combined use of albumin and lymphocyte counts have been mentioned by the Onodera's prognostic nutritional index [15]. Hypoalbuminaemia has been shown to correlate with ideal body weight, weight loss, body cell mass and poor prognostic in cancer patients [16]. In our cohort, hypoalbuminaemia associated with weight loss, low REE and low FBG. Patients with serum albumin levels <35 g/l had the 5-year survival rate of 26.7 %, compared to 57.8 % in patients with serum albumin > 40 g/l. Serum albumin was proved to be an independent significant prognostic factor by multivariate analysis even after adjusting for potential confounding factors. To assess the impact of albumin on different tumor stages, we divided patients into subgroups by Nstage (N0 vs. N1-4), and found hypoalbuminaemia as a worse survival factor in both N0 group (X 2 = 11.078, p = 0.004) and N1-4 group (X 2 = 7.236, p = 0.027). But the multivariate prognostic analysis for subgroups of N0 and N1-4 exhibited that serum albumin as a weaker prognostic value than other potential prognostic factors. It may be explained by the different clinical characteristics of each subgroup which may impact on the prognostic value.  As a particular nonspecific marker of systemic inflammation, an elevated NLR is hypothesized to be associated with poor survival in various solid tumors [17]. But for esophageal cancer, its effect on long-term outcome is still controversial [18,19]. Most of previous studies did not find the positive predictive values of NLR in patients with esophageal cancer [20,21]. In this study, there was no significant difference in survival as the NLR cutoff value taken as five because many previous literatures reported optimal cutoff value as five [22]. We also did not find the prognostic value of total lymphocyte counts in patients with esophageal cancer.
A few studies focused on the influence of BMI on postoperative outcomes in patients with EC and had shown contradictory results, varying from no differences in postoperative complications and mortality to a higher incidence of individual complications (such as respiratory complications and anastomotic leak) in high BMI patients [5][6][7]. In our study, we found no difference of the frequency of postoperative complications and mortality among different BMI patients. In addition, patients with a high BMI had a higher frequency of diabetes.
Smoking patients appeared to have a lower BMI. Although smoking is the preventable cause for esophageal cancer, but it was not observed to be a poor prognostic factor in our finding.
Previous studies had revealed contradictory results regarding the association between BMI and long-term outcome in esophageal cancer patients [4][5][6][7][23][24][25]. Most of these studies were conducted in western populations that have a high incidence rate of EAC, which occurs more frequently in patients with a high BMI. But the main type in Chinese populations is ESCC (67.9 % in our data). The population is prone to be lean (26.5 % with BMI < 20 and only 12.7 % with BMI > 25 in our data) compared to Western populations. Based on our data, compared with normal patients, we found a significantly worse OS in low BMI patients and a significantly better OS in high BMI patients on a univariate analysis. The prognostic effect of BMI seems to be more valuable on lymph node-negative patients than lymph nodepositive patients and proved to be independent prognostic indicators.
High blood glucose level makes patients more susceptible to certain postoperative complications such as surgical site infections, sepsis, myocardial ischemia and so on [26]. But no similar studies were reported concerning esophageal cancer. Based on our date, FBG did not affect the risk of all-cause mortality and postoperative complications. Our study found low FBG level and hypoalbuminaemia are inter-related. In the class of albumin <35 g/l, only 44.4 % (16/36) patients had the FBG > 90 mg/dL, but in the class of albumin >40 g/l, almost 76.1 % (105/138) patients had the FBG > 90 mg/dL, and both hypoalbuminaemia and low FBG level were proved to be worse survival factors. It is easy to measure and monitor FBG or serum albumin in clinical research, so both FBG and serum albumin can easily be used for survival prognosis, and researchers can choose the appropriate method according to their laboratory conditions. To assess the impact of FBG on prognosis in patients with esophageal cancer, both the univariate and multivariate analysis proved that low FBG level (≤90 mg/dL) was independently associated with poor OS. These finding is consistent with previous studies which suggested that hyperglycemia related microvessel changes may have a protective effect against neoplastic cell spread and metastasis in patients with malignant tumors such as non-small cell lung cancer and so on [27,28].
REE is believed to be elevated in several types of tumors, It has been hypothesized that increased energy expenditure may contribute to the development of malnutrition and weight loss [29]. Our study found patients with a low REE had significantly worse survival compared with high REE (5-year survival rate: 40.4 % vs 55.3 %, p = 0.010). In addition, there was no significant association of REE with postoperative complications and mortality. To our knowledge, it is the first demonstration of the prognostic value of estimated REE in postoperative patients with EC. Our study also observed a significant linear correlation between BMI and REE (p < 0.001), showing that patients with low BMI tended to have low REE. As the estimated REE largely depends on patients' weight and height, the worse prognosis for low REE patients may reflect that these patients are under-nourished and are likely to have poor prognosis as a result. In addition, patients with low REE tended to be associated with advanced T-stages, so we concluded that the advanced tumor stages may be another reason to explain the low REE patients' worse survival.
We also explore the relationship between REE and REE/kg. We found the effect of REE on patient survival was opposite to that of REE/kg. Low REE as well as high REE/kg was found to be a potential worse prognostic indicator. The existing hypothesis of REE is, the sum of the metabolic activities of internal organs, muscle, bone, and adipose tissue and can reflect patient's physiques and muscle volumes, but REE/kg may reflect the diffusion, consumption and metabolic rate of muscle. So patients with high REE/kg have been found to lose more weight, have lower BMI and have a worse survival, this finding is consistent with the hypothesis.
In this research, low FBG was significantly associated with low BMI, albumin and REE, a potential explanation might be proposed. Individuals with esophageal cancer commonly experience metabolic abnormalities. The abnormal state of insulin (hyperinsulinemia or insulin resistance) is one cause, and it promotes cancer growth and progression through its effects on the insulin and insulin-like growth factor pathways [30]. Meanwhile, diabetic microangiopathy render the vascular basal membrane less digestible by tumor cells, which may play a role in impeding neoplastic cell spread and metastasis.
Following are the main limitations in our study. First, this is a retrospective study and the sample size is not large. Second, all-cause mortality is used instead of disease-specific deaths, the latter is difficult to confirm. Third, we only chose R0 patients, so our results may not be suitable for all patients.

Conclusion
In conclusion, it is the first article to explore the prognostic significance of BMI, REE and FBG among patients with EC. FBG level ≤90 mg/dL was independently associated with poor survival for all patients. BMI and REE were important prognostic factors and the value was significant on lymph node negative patients. Therefore it is advisable that the combined assessment of BMI, REE and FBG can be used for a better preoperative assessment and prognostic evaluation in esophageal cancer patients.